Information and the aim of adjudication: truth or consequences?

• Author: Kaplow, Louis
• Position: III. Accuracy in Adjudication B. Damages and Ex Ante Behavior through Conclusion, with footnotes, p. 1337-1371 - Symposium: Festschrift in Honor of Richard Craswell

2. Damages and Ex Ante Behavior

   Next, we consider accuracy in the determination of damages. This Subpart addresses the consequences of greater accuracy for ex ante behavior, and the next examines the consequences with regard to the compensation of victims of harmful acts. The term "damages" evokes private civil litigation, but the analysis in this Subpart is equally applicable to fines (and that in the next Subpart applies as well to applications for compensation from the government, such as with disability claims under social insurance).

   Consider a simple situation in which, contrary to the analysis thus far, we can take for granted that it has been properly determined that a harmful act occurred. In addition, all harmful acts come before a tribunal, liability is strict, and the only question is the degree of harm the victim suffered. If actors ex ante have perfect information about the harm their acts cause, there are no system costs, and harm can be measured with perfect accuracy, then it is well known that setting damages equal to harm fully internalizes the externality caused by the harmful act and thereby induces socially optimal behavior. (65)

   Here, we will explore the implications of relaxing the assumptions about information. Starting with adjudication, let us compare two regimes. The initial system is of intermediate accuracy; specifically, the information is sufficient to enable the tribunal to form a correct estimate of average harm for the type of act. The contemplated reform creates a system that is entirely accurate--it measures the actual harm caused without error--but it requires that a cost be incurred. (66) For concreteness, one might suppose that after an initial, limited inquiry, the tribunal would have a general sense of the actual harm in a particular case (a middle-aged individual lost a limb) but that it would take substantial further effort, perhaps involving medical and occupational experts, in order to ascertain the particular victim's harm (lost wages from the particular injury in the victim's specific occupation). Our question is whether the incremental cost is justified in light of the social value of the additional accuracy, in this instance moving from a system that provides an unbiased but generally erroneous estimate of the truth to one that accurately determines the truth of the matter (harm) in every case.

   To analyze this question, one of Part I's lessons is central: differences in the outcomes of adjudication can influence prospective actors' ex ante behavior only if actors can anticipate the outcomes, whether because they immediately appreciate what those outcomes will be or because they are induced to incur the necessary expense up front to predict them. Accordingly, we will consider three cases, all of which are of practical importance in some settings: when prospective injurers cannot foresee the precise harm they will cause but only average harm; when they can perfectly and costlessly foresee the precise harm; and when they can do so but only if they incur a cost. A more accurate legal system, which produces the same truth regardless, has quite different consequences in each of these settings, and therefore the social value of accuracy differs significantly as well. (67)

   Our first case is where prospective injurers can predict average harm but cannot foresee the actual harm that would be suffered by the particular victim who might be injured. This case often seems to be realistic to a substantial degree. For example, careless driving will increase the likelihood of an accident, but just who would be hurt, by how much, and how that specific injury would affect the particular victim will be unforeseeable. At the time decisions about acts are made, individuals will often have (at best) a rough sense of the harm they might cause on average and also would find it essentially impossible through further research to generate precise knowledge of actual harm.

   In this case, greater accuracy in the measurement of damages has no consequences whatsoever for behavior and hence the system cost is a waste. That is, truth has no consequences (aside from consuming resources) and hence no social value. This conclusion at one level is obvious because the later refinement is, by assumption, impossible to foresee. To make the argument more complete, explicitly compare actors' incentives under the two regimes. Under the first, they understand that the legal system will set damages equal to average harm (equivalently, expected harm) and hence they will act if and only if their private benefit exceeds this amount. Under the second, they understand that the legal system will set damages equal to actual harm. However, they do not know what actual harm will be. Instead, their actions ex ante will be dictated by their expectations. When harm proves to be unusually high, they will pay more than the average; when it is low, they will pay less. What is their expected damages payment, viewed ex ante? Because damages equal actual harm in each case, expected damages will equal expected harm, that is, average harm. But this is precisely what they expected to pay under the inaccurate regime, so their behavior will be identical.

   Our second case is where prospective injurers can predict actual harm perfectly and costlessly. Although this may never literally be true, the scenario captures an important feature of reality in some settings. For example, in contract breach cases, the party contemplating breach may know a good deal about the harm its particular contractual partner would suffer. And back in our torts setting, some injurers will know to some extent certain more refined particulars about the harm their acts will cause. For simplicity, we are considering the pure case in which, in the course of their normal activity and thus at no additional cost, individuals can foresee actual harm precisely.

   Here, accuracy has consequences: it induces behavior better tuned to actual circumstances and thereby generates greater social welfare. Greater accuracy will be optimal when this welfare gain exceeds the cost of the more accurate legal system. Truth has consequences, desirable ones, but we nevertheless have a familiar cost tradeoff.

   To elaborate, under the system in which damages equal average harm, prospective actors commit harmful acts when their private benefit exceeds this level, as before. When harm is unusually high, they may commit acts despite the fact that their benefit is less than actual harm, and when harm is atypically low, they may abstain even though their benefit exceeds actual harm. An accurate legal system avoids both sources of social loss: because damages will be concomitantly higher in the former case, net socially harmful acts will be successfully deterred, and because damages will be lower in the latter case, net socially beneficial acts will not be chilled. The value of accuracy in this setting is the total of these gains from changes in behavior: the sum of the avoided net social harm for each of the former acts (which net will depend on the level of benefit for the particular act deterred) and the benefit from each of the latter acts now committed (which, again, will vary with the act). If this total exceeds the system costs of the more accurate regime, that regime will raise social welfare overall and hence be optimal. Note that accuracy--achieving truth in adjudication with regard to damages reflecting actual harm--tends to be more valuable overall when accuracy is less costly, when harm varies substantially across cases relative to the initial average estimate for the pertinent subclass, and when many prospective actors have private benefits in the range in which changing damage awards will induce changes in behavior. Only the second of these three factors relates directly to any notion of truth. (68)

   Our third case is where prospective injurers can predict the precise harm perfectly, but they must incur a cost to do so. Many settings would seem to have this feature to some extent. That is, actors would have some sense of the harm they might cause, but additional efforts might refine their estimates. We will confine attention to a pure version of this case that straddles the prior two cases: if no expenditure is made ex ante, individuals know only the average level of harm, but if an expenditure of some stated amount is made, they can predict harm perfectly.

   Here, accuracy may or may not have consequences for behavior, and, if it does, greater accuracy may or may not raise overall social welfare. The results for consequences, and for the social value, if any, of truth, are thus intermediate between those in our first two cases.

   One possibility is that the cost of becoming informed is sufficiently high that no one becomes informed ex ante. This is identical to case one, so greater accuracy in adjudication is a pure waste. The fact that it is possible to become informed ex ante but prohibitively expensive to do so does not help. Another possibility is that the cost, although positive, is not so high that no prospective actor would incur the cost. Who will do so, and what is the consequence of that? Actors with very high private benefits relative to average harm will tend to commit their acts, as before, without acquiring information about actual harm because such acquisition is costly but is unlikely to change their behavior. Even if they learn that harm is above average, their benefits are very high, so this knowledge would not ordinarily affect their privately optimal choice. Only in the unlikely event that harm is extremely high would they behave differently, and we are supposing that this is not sufficiently likely to justify the cost, which must be incurred for certain. Similar reasoning suggests that actors with very low private benefits relative to average harm will tend to abstain from action, as before, without acquiring information.

   Therefore, the only...